Adam

public class Adam

Optimizer that implements the Adam algorithm.

Adam optimization is a stochastic gradient descent method that is based on adaptive estimation of first-order and second-order moments.

According to Kingma et al., 2014, the method is "computationally efficient, has little memory requirement, invariant to diagonal rescaling of gradients, and is well suited for problems that are large in terms of data/parameters".

@see Kingma et al., 2014, Adam: A Method for Stochastic Optimization.

Constants

float	BETA_ONE_DEFAULT
float	BETA_TWO_DEFAULT
float	EPSILON_DEFAULT
String	FIRST_MOMENT
float	LEARNING_RATE_DEFAULT
String	SECOND_MOMENT

Inherited Constants

From class org.tensorflow.framework.optimizers.Optimizer

String

VARIABLE_V2

Public Constructors

	Adam(Graph graph) Creates an Adam optimizer
	Adam(Graph graph, float learningRate) Creates an Adam optimizer
	Adam(Graph graph, float learningRate, float betaOne, float betaTwo, float epsilon) Creates an Adam optimizer
	Adam(Graph graph, String name, float learningRate) Creates an Adam optimizer
	Adam(Graph graph, String name, float learningRate, float betaOne, float betaTwo, float epsilon) Creates an Adam optimizer

Public Methods

static <T extends TType> Op	createAdamMinimize(Scope scope, Operand<T> loss, float learningRate, float betaOne, float betaTwo, float epsilon, Options... options) Creates the Operation that minimizes the loss
String	getOptimizerName() Get the Name of the optimizer.
String	toString()

Inherited Methods

From class org.tensorflow.framework.optimizers.Optimizer

Op	applyGradients(List<GradAndVar<? extends TType>> gradsAndVars, String name) Applies gradients to variables
<T extends TType> List<GradAndVar<?>>	computeGradients(Operand<?> loss) Computes the gradients based on a loss operand.
static String	createName(Output<? extends TType> variable, String slotName) Creates a name by combining a variable name and a slot name
abstract String	getOptimizerName() Get the Name of the optimizer.
<T extends TType> Optional<Variable<T>>	getSlot(Output<T> var, String slotName) Gets the slot associated with the specified variable and slot name.
final Ops	getTF() Gets the Optimizer's Ops instance
Op	minimize(Operand<?> loss) Minimizes the loss by updating the variables
Op	minimize(Operand<?> loss, String name) Minimizes the loss by updating the variables

From class java.lang.Object

boolean	equals(Object arg0)
final Class<?>	getClass()
int	hashCode()
final void	notify()
final void	notifyAll()
String	toString()
final void	wait(long arg0, int arg1)
final void	wait(long arg0)
final void	wait()

Constants

public static final float BETA_ONE_DEFAULT

Constant Value: 0.9

public static final float BETA_TWO_DEFAULT

Constant Value: 0.999

public static final float EPSILON_DEFAULT

Constant Value: 1.0E-8

public static final String FIRST_MOMENT

Constant Value: "m"

public static final float LEARNING_RATE_DEFAULT

Constant Value: 0.001

public static final String SECOND_MOMENT

Constant Value: "v"

Public Constructors

public Adam (Graph graph)

Creates an Adam optimizer

Parameters

graph	the TensorFlow graph

public Adam (Graph graph, float learningRate)

Creates an Adam optimizer

Parameters

graph	the TensorFlow graph
learningRate	the learning rate

public Adam (Graph graph, float learningRate, float betaOne, float betaTwo, float epsilon)

Creates an Adam optimizer

Parameters

graph	the TensorFlow graph
learningRate	the learning rate
betaOne	The exponential decay rate for the 1st moment estimates. Defaults to 0.9.
betaTwo	The exponential decay rate for the 2nd moment estimates. Defaults to 0.999.
epsilon	A small constant for numerical stability. This epsilon is "epsilon hat" in the Kingma and Ba paper (in the formula just before Section 2.1), not the epsilon in Algorithm 1 of the paper. Defaults to 1e-8.

public Adam (Graph graph, String name, float learningRate)

Creates an Adam optimizer

Parameters

graph	the TensorFlow graph
name	the Optimizer name, defaults to "Adam"
learningRate	the learning rate

public Adam (Graph graph, String name, float learningRate, float betaOne, float betaTwo, float epsilon)

Creates an Adam optimizer

Parameters

graph	the TensorFlow graph
name	the Optimizer name, defaults to "Adam"
learningRate	the learning rate
betaOne	The exponential decay rate for the 1st moment estimates. Defaults to 0.9.
betaTwo	The exponential decay rate for the 2nd moment estimates. Defaults to 0.999.
epsilon	A small constant for numerical stability. This epsilon is "epsilon hat" in the Kingma and Ba paper (in the formula just before Section 2.1), not the epsilon in Algorithm 1 of the paper. Defaults to 1e-8.

Public Methods

public static Op createAdamMinimize (Scope scope, Operand<T> loss, float learningRate, float betaOne, float betaTwo, float epsilon, Options... options)

Creates the Operation that minimizes the loss

Parameters

scope	the TensorFlow scope
loss	the loss to minimize
learningRate	the learning rate
betaOne	The exponential decay rate for the 1st moment estimates.
betaTwo	The exponential decay rate for the 2nd moment estimates.
epsilon	A small constant for numerical stability. This epsilon is "epsilon hat" in the Kingma and Ba paper (in the formula just before Section 2.1), not the epsilon in Algorithm 1 of the paper.
options	Optional Optimizer attributes

Returns

• the Operation that minimizes the loss

Throws

IllegalArgumentException	if scope does not represent a Graph

public String getOptimizerName ()

Get the Name of the optimizer.

Returns

• The optimizer name.

public String toString ()